PyTorch and Detect.py get different results?

[alevangel]

Hello all,
I'm very new to the matter, I'm trying to use the YOLOv5 with PyCharm. In roder to do that i'm using the PyThorch Hub, but when I do a simple model(img) I get very different results from detect.py executed with a notebook.

I also configured the correct size of image (that in my case is 1028) and custom weights, this a snippet:

# Model
model = torch.hub.load(PATH+'yolov5', 'custom', path=PATH+'best.pt'
                       , force_reload=True
                       , source='local')
# Image
img = PATH+'test/images/66.jpg'

# Inference
model.conf = 0.55
results = model(img, size=1280, augment=True)

While this is what I do with detect.py:
!python detect.py --weights runs/train/yolov5m_results14/weights/best.pt --img 1024 --conf 0.55 --source /content/test/images

Actually I get different confidence levels and also different boxes, why this happens?

[glenn-jocher]

@alevangel you're running inference with different image sizes.

[alevangel]

Thank you so much... I don't why was struggling so much about it! D: In my biased mind were equal numbers!

[glenn-jocher]

@alevangel also augmented inference is very different, it's going to produce better results but take longer. The augment=True addition to your hub inference is doing this while detect.py is not augmented in your command.

[alevangel]

@glenn-jocher Uh I see, thanks! I thought it was augmented by default with detect.py, how to set it?

[glenn-jocher]

@alevangel see detect.py argarser for inference options:

python detect.py --augment

yolov5/detect.py

Lines 248 to 277 in de534e9

	def parse_opt():
	parser = argparse.ArgumentParser()
	parser.add_argument('--weights', nargs='+', type=str, default='yolov5s.pt', help='model.pt path(s)')
	parser.add_argument('--source', type=str, default='data/images', help='file/dir/URL/glob, 0 for webcam')
	parser.add_argument('--imgsz', '--img', '--img-size', nargs='+', type=int, default=[640], help='inference size h,w')
	parser.add_argument('--conf-thres', type=float, default=0.25, help='confidence threshold')
	parser.add_argument('--iou-thres', type=float, default=0.45, help='NMS IoU threshold')
	parser.add_argument('--max-det', type=int, default=1000, help='maximum detections per image')
	parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
	parser.add_argument('--view-img', action='store_true', help='show results')
	parser.add_argument('--save-txt', action='store_true', help='save results to *.txt')
	parser.add_argument('--save-conf', action='store_true', help='save confidences in --save-txt labels')
	parser.add_argument('--save-crop', action='store_true', help='save cropped prediction boxes')
	parser.add_argument('--nosave', action='store_true', help='do not save images/videos')
	parser.add_argument('--classes', nargs='+', type=int, help='filter by class: --class 0, or --class 0 2 3')
	parser.add_argument('--agnostic-nms', action='store_true', help='class-agnostic NMS')
	parser.add_argument('--augment', action='store_true', help='augmented inference')
	parser.add_argument('--visualize', action='store_true', help='visualize features')
	parser.add_argument('--update', action='store_true', help='update all models')
	parser.add_argument('--project', default='runs/detect', help='save results to project/name')
	parser.add_argument('--name', default='exp', help='save results to project/name')
	parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')
	parser.add_argument('--line-thickness', default=3, type=int, help='bounding box thickness (pixels)')
	parser.add_argument('--hide-labels', default=False, action='store_true', help='hide labels')
	parser.add_argument('--hide-conf', default=False, action='store_true', help='hide confidences')
	parser.add_argument('--half', action='store_true', help='use FP16 half-precision inference')
	opt = parser.parse_args()
	opt.imgsz *= 2 if len(opt.imgsz) == 1 else 1 # expand
	return opt